The records told the story backwards. Not chronologically — the files were organized by date, sequential, the research program's documentation culture producing a complete longitudinal record. But the story they told ran backwards from the conclusion that Im Byeongsoo had reached and never documented explicitly: the cultivation methodology didn't work.
Sora read through the night. The laptop that Dohyun's operational network had included with the apartment's provisions — clean, not connected to any network, the data from Eunji's drive loaded onto the local storage. Soojin asleep in the far bedroom, the ten-meter distance maintained even in sleep. The apartment's residential noise floor providing its continuous dampening cover.
The cultivation records documented four specimens. The designations were clinical: Subject C-1 through C-4. No names. The dehumanization was methodological, not intentional — the research program treated the specimens as biological architecture under study, and the documentation reflected the treatment.
Subject C-1: initiated twenty-three months before Sora's intake. An E-rank healer who'd sustained channel micro-fractures during a C-rank dungeon operation. The micro-fractures were the same structural precondition that Sora's Thornveil experience had escalated into full mutation. Im Byeongsoo's program identified C-1's damaged architecture as a potential cultivation candidate and began clockwise mana application at 0.34 THz — the same frequency used on Soojin.
The record documented C-1's development over sixteen months. The clockwise applications at biweekly intervals. The nodal development: a second node forming at regular geometric spacing — 180 degrees from the primary, the symmetric positioning that the cultivation methodology's clockwise input encouraged. Growth rate: approximately 1.5% per day during active cultivation.
At month twelve, C-1's second node reached 65% structural density. The symmetric geometry was stable. The clockwise cultivation was producing a dual-node architecture with regular spacing and no asymmetric interference.
At month fourteen, the second node's growth stalled. 72% density. The documentation noted: *Symmetric geometry approaching density ceiling. The regular 180-degree positioning creates minimal interference between nodes, resulting in no amplification factor. The architecture's effective output at 72% dual-node density does not exceed the parameter ceiling's threshold.*
No amplification. The symmetric geometry — the regular spacing that the clockwise cultivation produced — didn't create the interference pattern that Sora's asymmetric geometry generated. The cultivation's design flaw was geometric: by forcing symmetric development, the methodology eliminated the architectural feature that made mutation functional.
At month sixteen, C-1's second node began regressing. The documentation's assessment: *Without an amplification mechanism, the architecture lacks the metabolic justification for maintaining dual-node infrastructure. The System's parameter layer does not support dual-node healer-class architecture at standard amplitude levels. The developing node's metabolic cost exceeds its functional contribution.*
The architecture was rejecting the cultivation. Not damage. Not harm. The symmetric geometry simply didn't produce enough benefit to justify the cost. The body ran its own efficiency audit, found the second node was costing more than it delivered, and started taking it apart.
C-1's second node regressed to 40% density before the research program intervened with increased clockwise application amplitude. The intervention stabilized the regression but didn't restore growth.
The documentation's conclusion for C-1: *Symmetric cultivation ceiling reached. Architecture unable to progress past 72% without amplification mechanism.*
The same story repeated for C-2 and C-3. Different starting conditions — C-2 was a deliberate induction through controlled dungeon stress, C-3 was a natural micro-fracture case similar to C-1. Same methodology. Same clockwise cultivation. Same symmetric geometry. Same density ceiling between 68% and 74%. Same regression when the architecture rejected the non-functional node.
C-4 was different.
C-4's record began eleven months before Sora's intake. An E-rank healer who'd sustained channel damage during a B-rank dungeon operation — more severe than C-1 through C-3's initial conditions. The damage pattern included asymmetric micro-fractures: the fractures weren't uniformly distributed across the channel substrate but concentrated in a specific angular range.
Im Byeongsoo's documentation noted the asymmetry with clinical interest: *C-4's fracture pattern creates an irregular stress distribution in the channel substrate. The asymmetric damage may influence nodal development geometry if cultivation is initiated.*
The cultivation began with the same clockwise protocol. But C-4's second node developed differently. Instead of the 180-degree symmetric positioning that C-1 through C-3 produced, C-4's second node formed at 165 degrees — fifteen degrees offset from the symmetric position. The asymmetric fracture pattern had influenced the node's development geometry.
Fifteen degrees. The same offset that Soojin's developing second node carried.
The documentation tracked C-4's development with increasing intensity. The asymmetric offset was producing what the symmetric specimens hadn't: a detectable interference pattern between the two nodes. Small — the 15-degree offset generated minimal interference compared to Sora's sixth node's larger angular deviation — but present. Measurable. The beginning of the geometric exploit that Sora's architecture used at full scale.
At month seven, C-4's second node reached 30% density. The interference pattern was producing a measurable amplification factor: 1.05x. Barely above unity. But above.
Im Byeongsoo's documentation shifted at this point. The language changed from clinical observation to research objective: *C-4's asymmetric development must be evaluated against the cultivation protocol's design parameters. The asymmetric interference represents a deviation from the controlled cultivation methodology. Assessment: whether to correct the asymmetry or allow continued asymmetric development.*
To correct or to allow. The research program's decision point.
The next entry was the decision: *Correction initiated. Clockwise application amplitude increased to 0.38 THz (from 0.34 THz standard) with directional bias toward the asymmetric offset. The correction aims to rotate the developing node toward the 180-degree symmetric position.*
They'd tried to fix the asymmetry. Push the developing node from its natural 165-degree position toward the 180-degree position that the cultivation protocol demanded. Increase the clockwise application to force the geometry toward symmetry.
The correction's result, documented over the next four months: the developing node's offset decreased from 15 degrees to approximately 10 degrees. The interference pattern decreased proportionally. The amplification factor dropped from 1.05x to 1.02x. The architecture's functional benefit from the developing node decreased as the geometry was forced toward symmetry.
And the channel junctions — the connections between the primary node and the developing node that carried the mana flow between them — began to fracture.
The correction's clockwise bias created directional stress on junctions that had formed under asymmetric conditions. The junctions were built for the 165-degree geometry. Pushing the node toward 180 degrees meant the junctions were supporting a node that was moving away from the position they were designed to connect to. Like stretching a bridge while the foundations stay in place.
Junction fracture count over four months: three. From six original connections to three surviving.
C-4's current state, per the final entry in the cultivation record: developing second node at 30% density. Offset 10 degrees from symmetric. Three surviving junctions. Amplification factor: 1.02x. Growth: stalled. The architecture locked between the natural asymmetric development it had been building toward and the forced symmetric correction that the cultivation protocol had imposed.
Sora set the laptop down.
C-4 was Soojin.
The subject designation stripped of its human content. The clinical record that documented eleven months of clockwise cultivation as a research protocol and characterized the junction fractures as data points. The person in the other room of this apartment — sleeping, at ten meters, with her three surviving connections and her counterclockwise residual climbing — had been Subject C-4 in Im Byeongsoo's filing system.
The research program hadn't just failed to replicate Sora's mutation. It had taken a naturally developing asymmetric architecture — the 15-degree offset, the interference pattern, the beginning of the geometric exploit — and destroyed it. Pushed it toward a geometric configuration that didn't work, in the process fracturing the junctions that connected the developing node to the primary architecture.
The 15-degree offset had been the beginning of mutation. Not the same as Sora's architecture — a different geometry, a different angle, a different starting point. But the same principle: asymmetry producing interference, interference producing amplification, amplification producing the functional output that bypassed the parameter ceiling.
If Im Byeongsoo had left the asymmetry alone, Soojin's architecture might have continued its natural development. The second node might have reached a density where the 1.05x amplification factor grew. The interference pattern might have strengthened. The architecture might have found its own path around the ceiling, the way Sora's had — not through force, not through cultivation, but through the biological system's natural adaptation to constraint.
Instead, the research program had corrected the deviation. Forced the node toward symmetric positioning. Fractured the junctions. Stalled the growth. Locked the architecture in a state that was neither the natural mutation it had been developing toward nor the controlled cultivation it had been designed to become.
Sora looked at the bedroom door where Soojin slept.
The cultivation records' final data point: C-4's architecture at month eleven. Three junctions. Thirty percent density. 1.02x amplification. Growth stalled.
And in the margins of the record — not in the formal documentation but in Im Byeongsoo's personal annotation layer that Eunji's copy had preserved — a notation.
*C-4's natural asymmetric offset was the most promising development the cultivation program has produced. The 165-degree geometry mirrors the pre-System mutation documentation's descriptions of early-stage architectural adaptation. If the offset had been allowed to develop without correction, the specimen might have produced a viable asymmetric architecture. The correction was a mistake.*
A mistake. Documented in the personal annotations that never reached the formal record. Im Byeongsoo's private acknowledgment of what the research program's institutional framework required him to deny publicly: the natural development had been working, and the cultivation protocol had destroyed it.
Sora sat in the apartment's living area. The laptop's screen casting blue light into the pre-dawn darkness. The cultivation records' complete dataset downloaded and reviewed. The story that ran backwards from a conclusion that had never been formally reached: the way to produce healer-class mutation was to support the natural process, not to override it.
And the natural process that Soojin's architecture had been developing — the 15-degree asymmetric offset, the interference pattern, the counterclockwise adaptation — had been interrupted at 30% density with three surviving junctions and a suppression overlay that was slowly decaying.
The process wasn't dead. The cultivation records documented that: the counterclockwise residual at 8% was the architecture's minimum survivable natural resonance. The 0.27 THz frequency that Soojin's channel pathways had learned from Sora's four-second pulse was the architecture's communication channel for inter-healer resonance. The three surviving junctions, stressed and scarred, were still connected.
The process was damaged. Locked. Stalled. But the biological foundation was present.
The question was whether eleven days was enough time to reach the capability that could restart it.
Sora closed the laptop. Lay on the apartment's couch. The dampening routing cycling. The passive reception on standby. The sixth node at seventy-four percent, growing at 1.9% per day now that adequate nutrition had reached the metabolic system.
In the other room, Soojin's mana signature carried its quiet pattern. The counterclockwise residual at approximately 11.2%. The modulation cycle lengthening as the suppression decayed. The three junctions holding.
Thirty-two days to the threshold. Eleven to parity.
Twenty-one days of margin. And a set of cultivation records that explained exactly what had gone wrong, which meant they contained the negative space of what could go right.
Sleep arrived not like anesthesia this time. It arrived like the body making a decision — the metabolic systems assessing that the data had been absorbed, the clinical assessment had been completed, and the architecture's development required the restorative function that consciousness consumed.
She slept. The sixth node grew. The clocks ran.
Somewhere in Gwangju, Eunji's brother prepared for day one hundred and seven's rotation.
